Mistral Large 4 enters public preview with 1.05 trillion parameters
Mistral’s new multimodal MoE model enters public API preview with 1.05 trillion total parameters and a one-million-token context window. The current model documentation specifies 52 billion active parameters.
AT A GLANCE
Key points
- 1.05 trillion total; 52 billion active according to current documentation
- A 1.6-billion-parameter vision encoder
- One-million-token context window
- Public preview through the API
What is changing?
Mistral AI opened Mistral Large 4 to public preview on October 6. The multimodal Mixture-of-Experts model has 1.05 trillion total parameters, a 1.6-billion-parameter vision encoder and a one-million-token context window. The live model documentation checked on October 6 lists 52 billion active parameters, while earlier announcement-based coverage stated 49 billion. We corrected the technical figure to reflect the current documentation.
The model can be tried through the Mistral API as mistral-large-4, with agent use cases including structured output, function calling and document processing. Open weights are planned for late October. Combining text, images and long documents may simplify production workflows. The release remains a preview, so teams should evaluate quality, latency and cost on their own data before using it in critical workflows.
Explore the original sources
Source date: 6 October 2026.
Published by AI Radar: 6 Oct 2026, 17:07 (Istanbul, UTC+3). Updated: 7 Oct 2026, 00:05.
FOLLOW THE TOPIC
Topics and tags
Share your discovery.
Send this story to a friend.
Reader comments
What do you think about this development? Share your experience or question. Comments are published after editorial approval.
Your name/nickname and comment become public if approved. Read: Privacy notice.
Loading comments…